I apologize in advance if this issue does not stem from an engine bug - I am entirely out of options and am unable to further diagnose this issue alone.
Our game, Midnight Racing: Tokyo, is experiencing an issue in a specific place within the experience (Tokyo Bay Area). Upon joining, all players immediately crash - symptomatically similar to an unyielding loop - before the Roblox client memory usage spikes to 100% of available memory. In some cases, user’s entire systems are crashing. This is preventing myself and other developers for the game from having any ability to debug outside of Roblox log files - which do not indicate a game-related issue.
I have tested previous known-good versions of the game to no avail, even though we have not published a new version of Tokyo Bay Area in over two weeks and this issue has suddenly began in the last ~24 hours.
Testing in studio results in the same behavior. Moving the majority of our server-side scripts and only running a server-mode test results in the same behavior, including the client crashing in a non-recoverable manner and memory usage spiking to 100%.
Importantly, Tokyo Bay Area is already an exceptionally large place, with the map alone capable of using ~2500mb of memory. There are (tens of) thousands of assets which would be acquired by content provider upon joining.
Expected behavior
The place should not crash. Users should be able to join the game. Client should not consume 100% of system memory.
Just not sure, if this is the case, why it only seemed to begin occurring after a Roblox update rather than a game update… I had tried problem-solving by moving pretty much all of our server-side scripts to ServerStorage, but this didn’t help. I could definitely be overlooking something, though, as the place is pretty large with plenty of outdated and hidden scripts.
I only posted this here as, like I said, the game was fine until a (presumed) engine change.
edit: I’ll continue looking into things on my end, although this is difficult as available logs don’t indicate a game-issue and no “band aid” solutions are working at this time.
@dear_gee I’m also a developer for this game. I just turned off streaming enabled for the place, used the test tab to start a local server, and it did not crash.
I think it also might be helpful to know that the place file size is very close to the maximum file size allowed by Roblox.
We rolled out a change to streaming feature recently that could be the root cause of the issue. I’m going to rollback that update for place 17485826750 for the time being and will notify you with more details once we have a better understanding of the issue. Thanks for surfacing this.
@dear_gee I just tried testing it in a few different ways:
Live public server (crash)
Studio LAN server (crash)
Live private server (teleport stalls/fails, no error message)
might be worth mentioning that after about 5 minutes of waiting on the game to allocate/connect me to a server, my client lost connection to the lobby server
We also use a custom private server system since it’s a multi-place universe. Our “custom” private server system uses the TeleportService:ReserveServer api to allocate game servers to vip server users.
Before each test, I restarted both the Client and Studio. Testing with StreamingEnabled turned off still doesn’t crash the game on the studio LAN server, and it appears to initialize normally.
Not sure if it’s worth noting but the Universe ID does not match the Place ID for this game.
Universe ID: 1198533421
Place ID: 17485826750
Not sure if all this information is going to be relevant but I appreciate you helping us figure it out!
Hi there! It appears the current live game is still experiencing crashes with the rollback, assuming the rollback went live. As @bigcrazycarboy mentioned, the IDs for the game are:
Game - 3339374541
Universe - 1198533421
Place experiencing issue - 17485826750
Do we need to publish the place for the rollback to go live?
Additionally, the TeleportService issue appears unrelated to the assumed streaming issue and is on our end.
@EnumBobbert@bigcrazycarboy It seems that the issue is not related to the recent changes to streaming. I am going to share more details about the investigations in the private thread.